Recruitment of patients into clinical research studies is a persistent and well-documented challenge [1]. The number of clinical trials open for patient enrollment has increased dramatically over time, growing from 4000 protocols in 2000 to more than 134,000 in 2023, which has intensified the challenge of patient recruitment [2]. Approximately 80% of clinical trials are delayed or terminated early because of recruitment problems [3]. Poor patient recruitment is the top cause of clinical trial delays [4]. Failure to meet recruitment goals delays treatment advances, threatens internal validity, and raises concerns about the generalizability of results [3]. There is a clear need to improve recruitment practices.

The integration of electronically collected patient-entered data within clinical practices offers innovative approaches to aid research recruitment [5,6]. While patient-entered data can help identify potential study participants through eligibility screening and interest assessment, these methods present distinct challenges. Using electronic patient questionnaires to screen for study eligibility, while seemingly straightforward, increases the burden of patient response and can create frustration among patients uninterested in research participation [7]. This approach also requires ongoing technical resources to maintain and update screening criteria as studies evolve. More problematically, additional recruitment outreach to patients who have declined participation via questionnaire violates patient autonomy—a cornerstone of ethical research [7]. This practice risks eroding trust in the health care system.

A more nuanced approach involves identifying a patient attribute associated with a higher propensity for research participation. This strategy enables research teams to prioritize outreach efforts efficiently to those with a higher likelihood of participating while avoiding prescribed eligibility criteria or direct elicitation of their interest in research. This method reduces the risk of alienating patients and reduces patient burden of questionnaire completion.

The objectives of this study were to (1) identify patient-reported questions that reflect perceptions about participating in clinical research studies and (2) determine whether patient responses to these questions are predictive of interest in participating in a precision medicine research registry.

This was a mixed methods study that used an exploratory sequential design. It was conducted in 2 phases. Phase 1 was a qualitative analysis that informed the phase 2 quantitative study. In phase 1, cognitive interviews were conducted to identify attribute questions that would reflect the likelihood that patients would participate in a clinical research study. In phase 2, the relationship between patients’ responses to the questions, identified in phase 1, and their interest in learning more about an ongoing research study was assessed in patients seen in primary care clinics. The identified “research perception” questions were added to an electronic questionnaire set that patients routinely complete before primary care office visits. The set also included a question about interest in being contacted about an ongoing precision medicine registry (Figure 1).

Phase 1 of this study was reviewed by the Cleveland Clinic Institutional Review Board and considered preparatory for research. Participants in phase 1 received an information sheet inviting them to participate in the cognitive interviews, which outlined the study’s purpose, participation requirements, and voluntary nature of their involvement. Each participant received a US $25 Amazon gift card as compensation. Phase 2 received exemption from the Cleveland Clinic Institutional Review Board (IRB 19-465), which determined that informed consent was not required for either phase of the study. Project data were securely stored and analyzed on Cleveland Clinic servers, with access restricted to authorized study personnel.

The 32 participants who completed the cognitive interviews had a range of self-reported health conditions. Of the 32 participants, 13 (41%) were female, 13 (41%) were aged 65 years or older, and 11 (34%) had previously participated in a clinical research study (Table 1).

Data saturation was reached, with participants in the cognitive interviews voicing strong support for clinical research, recognizing its important role in advancing patient care advancement, and agreeing that it should be part of Cleveland Clinic’s mission (Table 2). Participants indicated that they preferred receiving information about research opportunities in person directly from their physicians, citing the trust-based relationship with their health care providers and their health care providers’ unique understanding of their medical needs. While participants also viewed emails, MyChart messages, and telephone calls as effective ways to learn about research eligibility, they wanted these to serve as secondary forms of communication after initial discussion with their physicians.

Participants held highly favorable views toward research, with 97% (31/32) agreeing or strongly agreeing that it should be part of the institution’s mission and 100% (32/32) affirming that research enhances patient care. Likewise, nearly all respondents indicated that they would consider participating in research if it benefited them (32/32, 100%) or others (31/32, 97%). Participants indicated that they feel it is important to help others, and many indicated feeling a sense of responsibility to give back; however, they would not participate in a clinical study to help others if they were physically or mentally harmed in the process. Of the 32 participants, 31 (97%) also expressed that it was important or very important to be informed about clinical research opportunities for which they were eligible. Furthermore, a few (2/32, 6%) said that they would be disappointed if their physician did not inform them about clinical studies for which they may be eligible candidates.

There was slightly less consensus on the significance of practical factors such as location, time commitment, and compensation (Table 3).

Participants found all questions to be clear and easy to understand, with no suggestions for additional questions. As the most effective question for gauging patients’ interest in research participation, they identified the following: “I would consider participating in a clinical research study if it could potentially help me.” This question, along with 2 other candidates, was chosen for piloting in clinical practice. The selection of these questions was based on participant feedback, the distribution of participant responses, and their perceived relevance. The additional 2 questions were the following: “I would consider participating in a clinical research study if it could potentially help others” and “How important do you feel it is for your healthcare provider to let you know of research trials for which you may be eligible?”

A total of 1077 patients completed the “research recruitment” question and at least 1 of the 3 “research perception” questions and were included in the study cohort (mean age 48.3, SD 16.3 years; 625/1065 female, 58.68%; 661/1005 White, 65.77%; Table 4). Mean PROMIS GH physical and mental health summary scores were 49.4, SD 8.8 and 49.6, SD 9.6, respectively, similar to the mean of the US general population. At least moderate depressive symptoms were indicated by 11% (106/936) of the patients. Moreover, 14.1% (148/1050) indicated that they had concerns about paying their bills, and 2.57% (27/1051) indicated that they had difficulty with transportation to their medical appointments.

Of the 1077 patients, 278 (25.8%) responded affirmatively to the “research recruitment” question, indicating that they were interested in being contacted about the precision medicine registry. Patients who responded affirmatively were older than those who responded no (mean 51.4, SD 15.6 years vs mean 47.2, SD 16.4 years), had worse self-reported physical and mental health conditions, were more likely to report depressive symptoms (38/252, 15.1% vs 68/708, 9.6%), and indicated more social needs (72/273, 26.4% vs 94/777, 12.1%).

The majority of the patients responded positively to the “research perception” questions: 75% (800/1067) agreed or strongly agreed that they would consider participating in a clinical research study if it could potentially help others, while 81.02% (858/1059) indicated that they would consider participating if it would help them (Figure 2). In addition, 72.96% (785/1076) of the respondents felt that it was important for their health care team to let them know of clinical research studies for which they may be eligible.

More positive responses to all 3 “research perception” questions were significantly associated with interest in the precision medicine registry. A dose-response relationship was demonstrated for the responses to each question (Table 5). For the question “I would consider participating in a clinical research study if it could potentially help others,” responses of “agree” and “strongly agree” were significantly associated with interest in the precision medicine registry (adjusted odds ratio 8.40, 95% CI 2.48-28.5 and adjusted odds ratio 17.6, 95% CI 5.08-61.1, respectively). The multivariable models incorporating demographic and health factors alongside “research perception” questions demonstrated superior discriminative ability compared to models using “research perception” questions alone, with C-statistic values ranging from 0.716 to 0.752 in adjusted models compared to values ranging from 0.628 to 0.650 in unadjusted models (Table 5). The highest discrimination was achieved by the model that included the “research perception” question “I would consider participating in a clinical research study if it could potentially help me” (C-statistic=0.752).

The “research perception” questions demonstrated high sensitivity but limited specificity (Table 6). For responses of “agree” or “strongly agree” or “important” or “very important,” sensitivity exceeded 85% across all questions. Similarly high negative predictive values (>85%) indicated that patients who responded negatively to these questions were unlikely to express interest in the precision medicine registry. However, the questions demonstrated low specificity (24%-31%) and positive predictive values (30%), indicating that positive responses were not strong predictors of actual interest in the precision study.

This study demonstrated that the response to 1 question added to electronic patient questionnaire sets about patient perceptions regarding research may help prioritize recruitment efforts to patients who are more likely to participate in clinical research studies.

There are several potential approaches to using patient-entered data to aid in research recruitment (Table 7), with each having distinct advantages and limitations. While methods that directly assess patient eligibility and interest by using screening questions can reduce the research team’s workload, practical constraints within clinical care settings often limit their feasibility and may compromise respect for patient preferences.

Capturing patients’ attitudes toward research offers a balanced solution that preserves autonomy and prevents the frustration that may occur when those who have explicitly declined contact about research are nonetheless approached. In addition, this approach is concise, straightforward to implement, and requires minimal ongoing maintenance.

The question “I would consider participating in a clinical research study if it could potentially help me” demonstrated superior performance metrics, with both a higher area under the curve and greater sensitivity compared to the alternative questions. However, its direct personal nature may make it less suitable for integration into general clinical questionnaires. Conversely, the question “How important do you feel it is for your healthcare provider to let you know of research trials for which you may be eligible” had the lowest C-statistic value and sensitivity but may have a better contextual fit within standard previsit questionnaires, potentially making it more practical for routine clinical implementation. Specificity was relatively low for all questions; in this context, optimizing sensitivity over specificity can be considered advantageous, as it minimizes the risk of missing potentially interested participants while accepting a higher rate of false positives.

The cognitive interviews revealed that patients prefer to hear about research opportunities directly from their physician. A preference for health care provider to patient discussion about research has been noted by others [24-26]. Incorporating “research perception” questions within the clinical workflow supports this preference by enabling physicians to prioritize recruitment efforts during their office visit based on patient responses indicating an interest in research. This approach is used by the Movement Disorder Section of the Neurological Institute at our health system, where patients answer the “research perception” question “I think it’s important to be told of research trials for which I may be eligible” alongside other clinical questions before their visit. In the experience of the authors, this screening method has proven effective in prompting clinicians to spend additional time discussing relevant studies with patients who express interest.

Our adjusted models, which contained a “research perception” question and additional variables, had improved discriminative capacity. If electronically implemented, they could allow a more precise estimation of patients’ interest in research. However, targeting participants based on demographic and health status characteristics could introduce greater patient selection bias, reducing a study’s generalizability. Adjusted models would also be more resource intensive to implement.

Recent advances in EHR systems, apart from patient-entered data functionality, have also streamlined the patient recruitment process for clinical research. Many now have query capabilities that enable research teams to efficiently identify potentially eligible patients by searching structured data fields for specific eligibility criteria [27]. This automation has reduced the need for developing custom patient questionnaires to screen patients for eligibility. Artificial intelligence applied to EHR records can also now screen patients for eligibility criteria [28,29]. In addition, EHR systems can now directly engage potential participants through patient portals, automatically notifying them of trials that match their clinical profile. Although only a small percentage of patients contacted this way enroll as participants, this approach is effective at increasing recruitment because of the large volume of patients contacted through patient portal messages [30-32]. Integrating this EHR-based recruitment method with the patient attribute question can create a synergistic approach to optimize patient enrollment. Using multiple recruitment methods can increase enrollment yield [33,34].

Although this study was conducted at a single institution, the proposed approach is readily adaptable to other health systems that routinely collect patient-reported data during clinical visits. As technology advances, patient-reported data collection is becoming increasingly commonplace [35]. Adding a single question that requires no score calculations to existing questionnaire sets is straightforward and minimally increases the burden of patient response.

Several important limitations should be noted. First, although participants in the cognitive interviews reported a wide range of health conditions, they had more education than the general population, and approximately one-third of the interviewees (11/32, 34%) had previously participated in a clinical research study. Their opinions may differ from those with different educational backgrounds and experience with clinical research. Second, our assessment focused solely on the relationship between “research perception” questions and interest in a precision medicine registry. Patient attitudes may vary substantially across different types of clinical research. Third, responses to research participation questions might differ significantly if patients were facing an actual serious medical diagnosis rather than a hypothetical scenario. Fourth, our methodology for maintaining screening logs limited our ability to track actual study participation in the precision medicine registry. Instead, we could only measure the relationship between “research perception” questions and patients’ initial interest in learning more about the study. Finally, relying on patient-entered questionnaires as a screening tool has inherent limitations. This approach would only capture data from patients who complete electronic questionnaires, potentially underrepresenting racial and ethnic minority groups or populations considered vulnerable who may have limited access to patient portals [36].

The ability to identify patients who have a greater likelihood of participating in research studies by assessing patients’ attitudes toward research is a helpful tool to aid patient recruitment into clinical research. This approach allows research teams to prioritize their recruitment efforts more effectively without relying on strict eligibility criteria. It provides a framework for more nuanced and respectful engagement in the recruitment process, ultimately serving both research efficiency and patient experience. The use of broad indicators rather than specific eligibility questions creates a more flexible and sustainable recruitment strategy that respects patient preferences while maximizing research participation opportunities. Further evaluation of this approach is warranted to assess the ability to use patient-reported questions on attitudes toward research for other types of clinical trials that vary by risk of adverse outcomes, level of patient burden, and health condition.